For TMDL studies where sediment is identified as the pollutant causing the water quality impairment, detailed process-based models are often avoided due to the high input data requirements. The required data simply do not exist or collecting the data is prohibitively expensive and/or time consuming.  Hydrologic models such as GWLF, SWAT, and AGNPS/CONCEPTS, include channel erosion sub-models. The channel erosion routines used in these models vary from highly empirical to predominately process-based. Little research has been done to compare the effectiveness of these models in predicting erosion of the stream bed and banks.

 

PROCEDURES:

Stroubles Creek in Blacksburg, Virginia was included on the 1996 303(d) impaired waters report for a benthic impairment. A TMDL study to address this impairment was completed in 2003; sediment was determined to be the primary stressor on the benthic community. Data developed for the TMDL study will be used to generate “typical” input parameters for the three watershed-scale models being evaluated: GWLF, SWAT, and AGNPS/CONCEPTS. The models will not be independently calibrated to avoid calibration bias when comparing model performance. Erosion data from a 500-m reach of Stroubles Creek collected using erosion pins, scour chains and topographic surveying will be used to evaluate model performance. Model output will be evaluated statistically using multiple comparison tests and regression analysis. A qualitative comparison of model usability will also be presented based on the time required for model simulation and complexity of input parameter requirements.